[Coin-lpsolver] CBC LP Solve Times

[John J Forrest]

Vishnu,

I agree that it sounds a long time to solve an LP.  It sounds as if you 
are referring to the initial solve of the LP relaxation of an IP and maybe 
some settings are not as good as they could be for an LP.

You could try using the stand-alone executable "clp" and just do clp 
filename and see if that is better.  Or you can send me a logfile or the 
mps file to look at.

John Forrest





"Vishnu Erramilli" <vishnu.erramilli at optessa.com> 
Sent by: coin-lpsolver-bounces at list.coin-or.org
01/12/2007 12:24 PM

To
coin-lpsolver at list.coin-or.org
cc

Subject
[Coin-lpsolver] CBC LP Solve Times






Hi,

I have been working with a problem with the following characteristics (as
given by the CBC pre-solver): 47333 rows, 139880 columns, 1714427 elements
and there is an issue related to LP solve times (using the clp solver):

1) It took 37703.8 seconds to solve the LP problem on CBC. It seems to be
taking too much time because I ran the same problem on two different
solvers and one took ~2000 seconds and the other took 30 seconds which
leads me to conclude that there is definitely some setting issues with my
current CBC executable.

2) I am using the cbcExampleSample2 exe to solve the problem.

Is there any reason why an LP could have such a drastic difference in
solve times? Is there a CBC executable that any on this discussion board
use that I could use to reduce the solve times?

I think there is definitely some issue if the LP is taking 10 hours to
solve. I would be thankful if anybody could point me in the right
direction as regards to settings/exe's of the CBC solver that could
optimize the solve times.

Thanks,

Vishnu


_______________________________________________
Coin-lpsolver mailing list
Coin-lpsolver at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/coin-lpsolver

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/clp/attachments/20070112/956b8a88/attachment.html>